Interdecadal climate variability in an idealized coupled ocean-atmosphere-sea-ice model is studied. The ocean component is a fully three-dimensional primitive equation model and the atmospheric component is a two-dimensional (2D) energy balance model of Budyko-Sellers-North type, while sea ice is represented by a 2D thermodynamic model. In a wide range of parameters the model climatology resembles certain aspects of observed climate. Two types of interdecadal variability are found. The first one is characterized by northward-propagating upper-ocean temperature anomalies in the northwestern part of the ocean basin and a westward-propagating, wavelike temperature pattern at depth. The other type has larger-scale temperature anomalies that propagate westward in both the upper and deep ocean, along the sea ice edge. Both types of oscillations have been found previously in similar models that do not include sea ice. Therefore, the oscillation mechanism does not depend on sea-ice feedbacks nor is it modified very much by the inclusion of sea ice. For some parameter values, the interdecadal oscillations are self-sustained, while for others they are damped. Stochastic-forcing experiments show that, in the latter case, significant interdecadal signals can still be identified in the time series of oceanic heat transport. The periods of these signals, however, do not closely match those identified in a stability analysis of the deterministic model when linearized about its steady state. The authors show that linearization around the actual climatology of the stochastically forced integrations provides a better match for some of the modes that were poorly explained when linearizing about the deterministic model's steady state. The main difference between the two basic states is in the distribution of climatological convective depth, which is affected strongly by intermittent atmospheric forcing.

Two novel statistical methods are applied to the prediction of transitions between weather regimes. The methods are tested using a long, 6 000-day simulation of a three-layer, quasi-geostrophic (QG3) model on the sphere at T21 resolution. The two methods are the k nearest-neighbor classifier and the random-forest method. Both methods are widely used in statistical classification and machine learning; they are applied here to forecast the break of a regime and subsequent onset of another one. The QG3 model has been previously shown to possess realistic weather regimes in its Northern Hemisphere and preferred transitions between these have been deter- mined. The two methods are applied to the three more robust transitions; they both demonstrate a skill of 35–40% better than random and are thus encouraging for use on real data. Moreover, the random-forest method allows, while keeping the overall skill unchanged, to efficiently adjust the ratio of correctly predicted transitions to false alarms. A long-standing conjecture has associated regime breaks and preferred transitions with distinct directions in the reduced model phase space spanned by a few leading empirical orthogonal functions of its variability. Sensitivity studies for several predic- tors confirm the crucial influence of the exit angle on a preferred transition path. The present results thus support the paradigm of multiple weather regimes and of their association with unstable fixed points of atmospheric dynamics.

Two novel statistical methods are applied to the prediction of transitions between weather regimes. The methods are tested using a long, 6 000-day simulation of a three-layer, quasi-geostrophic (QG3) model on the sphere at T21 resolution. The two methods are the k nearest-neighbor classifier and the random-forest method. Both methods are widely used in statistical classification and machine learning; they are applied here to forecast the break of a regime and subsequent onset of another one. The QG3 model has been previously shown to possess realistic weather regimes in its Northern Hemisphere and preferred transitions between these have been determined. The two methods are applied to the three more robust transitions; they both demonstrate a skill of 35—40% better than random and are thus encouraging for use on real data. Moreover, the random-forest method allows, while keeping the overall skill unchanged, to efficiently adjust the ratio of correctly predicted transitions to false alarms.

A long-standing conjecture has associated regime breaks and preferred transitions with distinct directions in the reduced model phase space spanned by a few leading empirical orthogonal functions of its variability. Sensitivity studies for several predictors confirm the crucial influence of the exit angle on a preferred transition path. The present results thus support the paradigm of multiple weather regimes and of their association with unstable fixed points of atmospheric dynamics.

A statistical learning method called random forests is applied to the prediction of transitions between weather regimes of wintertime Northern Hemisphere (NH) atmospheric low frequency variability. A dataset composed of 55 winters of NH 700-mb geopotential height anomalies is used in the present study. A mixture model finds that the three Gaussian components that were statistically significant in earlier work are robust; they are the Pacific North America (P N A) regime, its approximate reverse (the reverse P N A, or RN A), and the blocked phase of the North Atlantic Oscillation (BN AO). The most significant and robust transitions in the Markov chain generated by these regimes are P N A -> BN AO, P N A -> RN A and BN AO -> P N A. The break of a regime and subsequent onset of another one is forecast for these three transitions. Taking the relative costs of false positives and false negatives into account, the random-forests method shows useful forecasting skill. The calculations are carried out in the phase space spanned by a few leading empirical orthogonal functions of dataset variability. Plots of estimated response functions to a given predictor confirm the crucial influence of the exit angle on a preferred transition path. This result points to the dynamic origin of the transitions.